Advances in AI pose growing threats to human life and health. State-of-the-art models have now proven capable of designing new viruses and toxins, illustrating that risks that were once considered only theoretical are now beginning to materialise across different domains of health. AI capabilities are progressing rapidly, creating a need for commensurate policy action to prevent population-scale harm. Karim's research examines how these advances may interact with the broad range of factors that mediate health, and the policy interventions that may consequently be required to safeguard public health as the technology continues to develop.